Cloud computing

Pier Siebesma (TU Delft/KNMI)

Subject: High Resolution Modeling of Deep Cumulus Convection

“My research is about deep convection. You can think of the tallest and widest clouds that occur in the atmosphere – say several kilometres wide and 10 kilometres high. What we are trying to investigate involves two issues that have a large effect on the formation and the development of these clouds.

The first half of my PhD has been about the influence of the layer below the cloud. The results were surprising! What we discovered is that it is not so much the magnitude of the extremes that determines whether a cloud will rise and become a very tall cloud, rather it is mainly the fact that the bigger clouds are already wider at cloud base.

Implementing this in a weather model is a process that takes quite some time. This work is in a very early stage. One of the things that have happened is that people just put in one very simple model for the organization of clouds in a weather model.

Weather models typically look at a scale of ten kilometres or more, which is coarser than individual clouds. We run computer simulations with a resolution of about 100 meters, and time steps of a few seconds, wherein you can actually see the individual clouds develop. Our simulations can also be thought of as a virtual laboratory for clouds.

Deep convective clouds are the clouds that occur in the tropics, also referred to as ‘cloud number 9’; the cumulonimbus cloud. This cloud is associated with heavy rainfall and wind gusts, so it’s very important to have the behaviour of these clouds modelled accurately in a weather model.

One of the funny things is that this often goes wrong in weather models. Weather models typically have a bias, a tendency to give this type of convection too early in the day. We’re trying to see if we can develop a model with which we can actually account for the gradual development of the clouds.

The second thing I’m looking at is the influence of the atmosphere above this layer, wherein the clouds form. For example, when the atmosphere is relatively dry around the cloud, it cannot become very tall, because it loses its moisture very rapidly. The influence of the air surrounding the cloud is much easier to implement than that of the air below. This makes it exciting, because sometimes with a very small tweak in defining the parameters, you can really improve the performance of a weather model.”

“It was one of my finest moments,” recalls Karel van Dalen (PhD student at Civil Engineering and Geosciences). “I had calculated the propagation of sound waves through porous stone and had predicted the waveforms at its surface. When I did the experiment some time later in the laboratory of the university in Leuven, the exact same waveforms appeared on the screen.” It confirmed to him that he was on the right track, and that he’d finally mastered the complex mathematics of how sound waves travel along the surface of and through porous material.

The practical applications of such knowledge lie in oil and gas exploration, but also in detecting groundwater or the examining of bones. Typically the damping or propagation speed is measured to assess certain material properties.  However, if one not only wants to find out how much gas is contained in the stone (the porosity) but also how mobile it is (the permeability), one must analyse the full waveform instead of just its speed and damping. Once Van Dalen had figured out how sound waves propagate, he could start on the inverse problem: given a particular waveform, what are the material properties causing it? He approached this by calculating the difference between measurement and model, and subsequently minimising the result.

The trouble was: this gave him not a single solution but rather a range of possible values for the gas content and mobility that fit the bill. Repeating the analysis for another type of wave (shear instead of compression), produced another range of solutions. However, the overlap of the two pretty precisely pinned down the values for porosity and permeability.

Until now, most of this work has been lab-based and computer-calculated. But in a bore hole the same technique could be used to map the permeability of the geological surroundings up to distance of 40 metres.
Van Dalen has recently published his findings in Geophysical Research Letters and will soon inform the Dutch Petrophysical Society of his findings.  

K.N. van Dalen, ‘Multi-component acoustic characterization of porous media’, 7 March 2011, PhD supervisors Professor Kees Wapenaar, Professor David Smeulders and Dr Guy Drijkoningen.